Find Calibration Parameter

This function finds the calibration parameter for type I error control in empirical Bayes power prior methods. It is based on the code in Nikolakopoulos et al, 2018, "Dynamic borrowing through adaptive power priors that control type I error". We just renamed the variables to be more explicit. The function finds the calibration parameter by the same bisection search as the original, but evaluates the type I error at each step exactly rather than by simulation. The result is deterministic.

Usage

findCalibrationParameter(
  n_iter = 1e+06,
  source_sample_size_per_arm,
  target_sample_size_per_arm,
  source_treatment_effect_estimate,
  desired_tie = 0.065,
  significance_level = 0.05,
  target_data_sampling_variance,
  source_data_sampling_variance,
  tolerance = 1e-04,
  theta_0 = 0
)

Arguments

n_iter

Formerly the number of simulated target estimates used to estimate the type I error. The type I error is now integrated exactly by adaptive_power_prior_type_I_error(), so this argument is ignored. It is retained because existing method configurations still supply it.

source_sample_size_per_arm

Sample size per arm in the source study

target_sample_size_per_arm

Sample size per arm in the target study

source_treatment_effect_estimate

Treatment effect estimate in the source study

desired_tie

Desired type I error rate

significance_level

Significance level for hypothesis testing

target_data_sampling_variance

Sampling variance of the target study data

source_data_sampling_variance

Sampling variance of the source study data

tolerance

Tolerance for convergence of the estimation

theta_0

True mean for type I error computation

Value

A matrix containing the calibration parameter and other related values